What challenges did Mortgage Advice Bureau face before AI implementation?

The complaints resolution team needed faster initial categorization, prioritization, and summarization of inbound customer complaint emails. Manual handling could delay initial responses by up to four days, which risked slower customer service and escalation. MAB needed a more consistent workflow for complaint intake while meeting FCA expectations.

What was the impact of the AI solution?

Automated ingestion happens within about 30 seconds of receipt. Initial response times were reduced by 75%.

Source

Proof of conceptEvidence: Medium65/100

Mortgage Advice Bureau and BJSS accelerate complaints process using Amazon Bedrock (AI on AWS)

Use case typeLegal document summarizationUpdated Jun 13, 2026

Mortgage Advice Bureau (MAB) is a UK mortgage intermediary that wanted to improve how it processed customer complaints and avoid delays before escalation to the Financial Ombudsman. MAB and AWS Partner BJSS built an AI solution on AWS that automatically ingests complaint emails, extracts required fields with Claude 3 Sonnet via Amazon Bedrock, summarizes and categorizes the complaint, identifies the relevant case context, and proposes an initial response. The system uses AWS Lambda for event-driven processing, Amazon S3 for storage, and Amazon RDS for relational data. The proof of concept was built in about five weeks.

Organization: Mortgage Advice Bureau
Industry: Finance
Location: United Kingdom
Published: June 2026

Reported outcomes

Time: −75%

Time & speed

Automation: Approximately 30 secondsTime: 10 minutes

Catalog median for time & speed deployments: −50% across 295 reported metrics. Compare benchmarks →

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Quantified impact: 30 seconds

AWS Partner SuccessJun 8, 2026Customer storyInferred claimMedium evidence strength

Automated ingestion happens within about 30 seconds of receipt.

Normalized claim

Time: 10 minutes

AWS Partner SuccessJun 8, 2026Customer storyInferred claimMedium evidence strength

Customers can receive a human-agent response in as little as 10 minutes.

Normalized claim

Time: 75% decrease

AWS Partner SuccessJun 8, 2026Customer storyInferred claimMedium evidence strength

Initial response times were reduced by 75%.

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Mortgage Advice Bureau
Provider: AWS
Maturity: PoC
Linked source: AWS Partner Success

The proof of concept was built in about five weeks

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 4

1Complaint triage
2Document summarization
3Workflow automation

BJSS worked with MAB to build an AI solution on AWS to automate complaint ingestion and triage.
The solution uses Claude 3 Sonnet through Amazon Bedrock to extract required fields from complaint emails.
It automatically summarizes, prioritizes, and categorizes complaints, identifies the relevant case or prior context, and drafts an initial response for the agent.
The architecture also includes AWS Lambda, Amazon S3, and Amazon RDS.

Technologies

Amazon Bedrock Claude 3 Sonnet AWS Lambda Amazon S3 Amazon RDS

Automated ingestion happens within about 30 seconds of receipt.
Initial response times were reduced by 75%.

Architecture

Complaint emails are ingested automatically; Claude 3 Sonnet via Amazon Bedrock extracts required fields; AWS Lambda handles event-driven processing; Amazon S3 stores documents; Amazon RDS stores relational complaint data.

Sources & evidence1

Evidence: Medium65/100Evidence strength